Structural basis of ligand binding modes at the neuropeptide Y Y1 receptor.

Neuropeptide Y (NPY) receptors belong to the G-protein-coupled receptor superfamily and have important roles in food intake, anxiety and cancer biology 1,2 . The NPY-Y receptor system has emerged as one of the most complex networks with three peptide ligands (NPY, peptide YY and pancreatic polypeptide) binding to four receptors in most mammals, namely the Y1, Y2, Y4 and Y5 receptors, with different affinity and selectivity 3 . NPY is the most powerful stimulant of food intake and this effect is primarily mediated by the Y1 receptor (Y1R) 4 . A number of peptides and small-molecule compounds have been characterized as Y1R antagonists and have shown clinical potential in the treatment of obesity 4 , tumour 1 and bone loss 5 . However, their clinical usage has been hampered by low potency and selectivity, poor brain penetration ability or lack of oral bioavailability 6 . Here we report crystal structures of the human Y1R bound to the two selective antagonists UR-MK299 and BMS-193885 at 2.7 and 3.0 Å resolution, respectively. The structures combined with mutagenesis studies reveal the binding modes of Y1R to several structurally diverse antagonists and the determinants of ligand selectivity. The Y1R structure and molecular docking of the endogenous agonist NPY, together with nuclear magnetic resonance, photo-crosslinking and functional studies, provide insights into the binding behaviour of the agonist and for the first time, to our knowledge, determine the interaction of its N terminus with the receptor. These insights into Y1R can enable structure-based drug discovery that targets NPY receptors.

NPY is a highly abundant nen class="Chemical">uropeptide in the central nervous system[7]. The first characterized NPY receptor Y1R is widely expressed in a variety of tissues and involved in regulation of many physiological functions, related to obesity[8] and cancer[9]. To better understand the ligand binding behavior of NPY receptors and provide a basis for drug discovery, we solved crystal structures of Y1R in complex with two structurally diverse antagonists, UR-MK299, an argininamide with high Y1R selectivity[10], and BMS-193885, which displays anorectic activity in animal models[6] (Fig. 1 and Extended Data Table 1). To facilitate structure determination, an engineered Y1R construct was designed (see Methods).

Figure 1

Structures of Y1R–UR-MK299 and Y1R–BMS-193885 complexes

a, Structure of Y1R–UR-MK299 complex. The receptor is shown in brown cartoon representation. UR-MK299 is shown as spheres with yellow carbons. b, Structure of Y1R–BMS-193885 complex. The receptor is shown in green cartoon representation. BMS-193885 is shown as spheres with pink carbons.

Extended Data Table 1

Data collection and refinement statistics

	Y1R–UR-MK299	Y1R–BMS-193885
Data Collection*
Space group	P21	C2221
Cell dimensions
a, b, c (Å)	37.8, 100.7, 83.2	76.9, 126.8, 170.3
α, β, γ (°)	90.0, 98.8, 90.0	90.0, 90.0, 90.0
Resolution (Å)	50.0-2.7 (2.8-2.7)†	50.0-3.0 (3.1-3.0)
Rmerge (%)	23.7 (83.9)	16.6 (93.7)
l/σl	5.01 (0.71)	5.30 (1.00)
Completeness (%)	93.0 (87.0)	92.4 (79.2)
Redundancy	7.4 (2.3)	3.5 (2.6)
Refinement
Resolution (Å)	50.0-2.7	50.0-3.0
No. reflections	15,818 (758)	15,600 (797)
Rwork/Rfree (%)	22.4/26.6	22.4/24.9
Number of atoms
Protein	3,791	3,654
Ligand	45	43
Overall B values (Å2)
Protein	88.3	108.0
Ligand	66.4	81.0
R.m.s. deviations
Bond lengths (Å)	0.009	0.009
Bond angles (°)	1.01	1.00

Diffraction data from 47 Y1R–UR-MK299 crystals and 33 Y1R–BMS-193885 crystals were used to solve the structures.

Numbers in parentheses refer to the highest-resolution shell.

Within the β-branch of class A GPCRs, to which n class="Gene">NPY receptors belong, the structures of four receptors, namely the neurotensin receptor NTS1[11], the OX1 and OX2 orexin receptors[12,13] and the endothelin ETB receptor[14], are determined to date. These structures reveal distinct differences of ligand binding modes between different receptors, suggesting that more structural information is needed to develop any consensus about the ligand recognition mechanisms for this GPCR subfamily. The Y1R structure shares a canonical seven transmembrane helical bundle (helices I-VII) with the other known GPCR structures (Fig. 1 and Extended Data Fig. 1a, b). The Y1R–UR-MK299 and Y1R–BMS-193885 complexes are structurally similar with Cα root-mean-square deviation (r.m.s.d.) of 0.75 Å within the helical bundle, and both exhibit inactive conformations with helix VI adopting a similar inward conformation as that in the other inactive GPCR structures. UR-MK299 binds to Y1R in a cavity within the helical bundle bordered by helices III, IV, V, VI and VII (Fig. 2a, b). The diphenylmethyl moiety of the antagonist interacts with a hydrophobic cluster formed by F2826.54, F2866.58 and F3027.35 (superscript: Ballesteros-Weinstein nomenclature[15]) on helices VI and VII of Y1R. The critical role of this hydrophobic patch in recognizing the argininamide-type Y1R antagonist was confirmed by the NPY-induced inositol phosphate (IP) accumulation of Y1R inhibited by UR-MK299 and several related Y1R antagonists, BIBP3226, BIBO3304, UR-HU404 and UR-MK289 (Extended Data Fig. 1e-i). The mutation F3027.35A abolishes the antagonistic activity for all these antagonists, while a 2-5-fold decreased antagonistic effect of all tested antagonists was observed for F2866.58A (Fig. 3a-c, Extended Data Fig. 2 and Extended Data Table 2).

Extended Data Figure 1

Crystal packing and structural features of Y1R and chemical structures of Y1R ligands

a, b, Crystal packing of Y1R–UR-MK299 (a) and Y1R–BMS-193885 (b) complexes. Y1R is shown in cartoon representation and coloured brown and green in the Y1R–UR-MK299 and Y1R–BMS-193885 complexes, respectively. The T4L fusion is shown in grey cartoon representation. UR-MK299 and BMS-193885 are displayed as yellow and pink spheres, respectively. c, Cutaway view of UR-MK299 binding pocket in Y1R. The receptor is shown in brown cartoon and surface representations. The ligand is shown as yellow sticks. d, Comparison of Y1R between the Y1R–UR-MK299 crystal structure (brown) and Y1R–NPY model (green). Side chains of Q1203.32 and W2766.48 are shown as sticks. R35-Y36 of NPY is displayed as cyan sticks. The hydrogen bond between Q1203.32 and Y36 of NPY is shown as a green dashed line. e-j, Chemical structures of argininamide Y1R antagonists, BIBP3226 (e), UR-HU404 (f), UR-MK299 (g), BIBO3304 (h), UR-MK289 (i) and UR-MK136 (j). k, Chemical structure of BMS-193885. l, Scaffold of NPY C-terminal residues R35 and Y36. Key differences between R35-Y36 of NPY and UR-MK299 are chirality of the arginine derivative and alteration of bond connectivity leading to the hydroxyphenyl group.

Figure 2

Ligand-binding pocket of Y1R for UR-MK299 and BMS-193885

a, Binding pocket for UR-MK299. The receptor is shown in grey cartoon representation. UR-MK299 (yellow carbons) and receptor residues (dark brown carbons) involved in ligand binding are shown as sticks. Salt bridge and hydrogen bonds are shown as red and green dashed lines, respectively. b, Schematic representation of interactions between Y1R and UR-MK299 analysed by LigPlot+ (ref. 30). The stick drawing of Y1R residues is coloured dark brown. c, Binding pocket for BMS-193885. BMS-193885 (pink carbons) and receptor residues (green carbons) involved in ligand binding are shown as sticks. d, Schematic representation of interactions between Y1R and BMS-193885 analysed by LigPlot+ (ref. 30). The stick drawing of Y1R residues is coloured green.

Figure 3

IP accumulation assays

a-i, NPY-induced IP accumulation of wild-type (WT) and mutant Y1Rs in absence of antagonist or in presence of BIBP3226 (10−5 M), BIBO3304 (10−6 M), UR-HU404 (10−7 M), UR-MK289 (10−5 M) or UR-MK299 (10−7 M). EC50 values of NPY (black) and EC50 ratios (EC50(NPY+antagonist)/EC50(NPY)) for antagonists (coloured) are given in the upper left corner for each plot. A reduced EC50 ratio of mutant compared to the wild-type receptor was interpreted as important for the respective antagonist. nd, not determined. j-m, Complementary mutagenesis assays of [N30]NPY with I293ECL3N (j, k) and [A33]NPY with N2997.32A (l, m). EC50 shifts (EC50(mutant)/EC50(WT)) are given in the upper left corner for each plot. A reduced EC50 shift of NPY analogue/Y1R mutant compared to NPY/Y1R mutant was interpreted as no further loss of function and a direct interaction between both positions. At least two (n) independent experiments were performed in technical duplicate. If n>2, data are shown as mean ± s.e.m. If n=2, data shown are from a representative experiment. See Extended Data Table 2 for detailed statistical evaluation.

Extended Data Figure 2

Expression of wild-type and mutant Y1Rs in transiently transfected COS-7 cells

a, Live cell fluorescence microscopy verifies all Y1R variants to be properly folded and exported to the cell membrane like the wild-type receptor. Nuclei stained with Hoechst33342, bar equals 10 μm. Pictures are representative of two independent experiments with similar results. b, Total expression level was determined by fluorescence reading and confirmed expression similar to the wild type. Transfection of only 50% or 25% of the DNA amount (with total DNA amount held constant by empty vector), respectively, led to a proportial decrease of fluorescence, and thus, expression level. Data represent means ± s.e.m of three to five independent experiments performed in technical triplicate (see source data for sample size n of each mutant). c, Estimation of the receptor reserve in functional IP accumulation assays. Transfection of half of the vector encoding receptor (with constant total DNA amount including chimeric G protein, cf. a) still produces maximum signal, while further reduction results in signal loss at comparable potency. Thus, there is only a small receptor reserve in the functional readout, allowing potency alteration to be directly related to compromised ligand binding. Data represent means ± s.e.m of three independent experiments performed in technical duplicate.

Extended Data Table 2

IP accumulation assays of wild-type and mutant Y1Rs for NPY and antagonists

Mutants	NPY		NPY/BIBP3226 (10−5 M)*				NPY/BIBO3304 (10−6 M)				NPY/UR-HU404 (10−7 M)				NPY/UR-MK289 (10−5 M)				NPY/UR-MK299 (10−7 M)
	EC50 (nM)†(pEC50 ± SEM)	n‡	EC50 (nM)(pEC50 ± SEM)	Ratio[§]	Kb(nM)‖	n	EC50 (nM)(pEC50 ± SEM)	Ratio	Kb(nM)	n	EC50 (nM)(pEC50 ± SEM)	Ratio	Kb(nM)	n	EC50 (nM)(pEC50 ± SEM)	Ratio	Kb(nM)	n	EC50 (nM)(pEC50 ± SEM)	Ratio	Kb(nM)	n
Wild type	1.7(8.78 ± 0.03)	20	463(6.43 ±0.04)	272	36.9	14	175(6.76 ± 0.04)	103	9.8	16	1,099/126#(5.96 ± 0.04)/(6.90 ± 0.07)	646/74#	0.16/0.14#	14	228(6.64 ± 0.04)	134	75.2	15	250(6.60 ± 0.06)	147	0.69	10
Y1002.64A	562;404	2	nd	nd	nd	2	nd	nd	nd	2	nd	nd	nd	2	nd	nd	nd	2	/	/	/	/
Q1203.32H	12; 15	2	nd	nd	nd	2	nd	nd	nd	2	>50,000; >50,000	>3,500	<0.03	2	>50,000>50,000	>3,500	<2.86	2	>50,000>50,000	>3,500	<0.03	2
Q1203.32N	3.8(8.42 ± 0.07)	4	836(6.08 ± 0.06)	220	45.7	3	414(6.38 ± 0.09)	109	9.3	3	3,744(5.43 ± 0.10)	985	0.10	3	325(6.49 ± 0.08)	86	118	3	743(6.13 ±0.06)	196	0.51	3
I1243.36A	6.0(8.22 ± 0.08)	3	1,877(5.73 ± 0.06)	313	32.1	4	306(6.52 ± 0.06)	51	20.0	4	1,588(5.80 ± 0.06)	265	0.38	4	637(6.20 ± 0.08)	106	95.2	4	208(6.68 ± 0.09)	35	2.94	4
Q2195.46A	23(7.63 ± 0.05)	7	2,732(5.56 ± 0.08)	119	84.8	4	1,156(5.94 ± 0.07)	50	20.4	3	>10,000/1,711#; 2,197#	>435/74#; 96#	< 0.23/0.14#; 0.11#	3/2#	5,518(5.26 ±0.12)	240	41.8	3	1146; 793	50; 34	2.05; 3.0	2
W2766.48A	3.8(8.42 ± 0.06)	5	246(6.61 ± 0.12)	65	156	3	84; 60	22; 16	47; 68	2	490; 254	129; 67	0.8; 1.5	2	1,445; 1505	380; 396	26; 25	2	140; 87	37; 23	2.8; 4.6	2
T2806.52A	2.6(8.58 ± 0.08)	4	141(6.85 ± 0.08)	54	189	4	124(6.91 ± 0.05)	48	21.3	4	871(6.06 ± 0.08)	335	0.30	4	278(6.56 ± 0.07)	107	94.3	3	306(6.51 ± 0.06)	118	0.86	3
N2836.55A	900(6.05 ± 0.06)	7	1,148(5.94 ± 0.12)	1	nd	3	1,036(5.98 ±0.14)	1	nd	3	7,622(5.12 ± 0.14)	8	14.3	3	1,193(5.92 ± 0.14)	1	nd	3	495; 571	1	nd	2
F2866.58A	4.5(8.35 ± 0.04)	7	491(6.31 ± 0.05)	109	92.6	3	118(6.93 ± 0.09)	26	40.0	3	1,553(5.81 ± 0.05)	345	0.29	3	126(6.90 ± 0.07)	28	370	3	180(6.74 ± 0.16)	40	2.56	3
D2876.59N	260(6.59 ± 0.05)	7	900; 748	3.5; 2.9	4065; 5328	2	341; 142	1	nd	2	>20,000; 1,571#(5.80 ± 0.12)	>77/6#	<1.3/2.0#	2/3#	729; 504	2.8; 1.9	5,543; 10,655	2	193; 177	1	nd	2
F3027.35A	4.2(8.38 ± 0.09)	3	16; 29	4; 7	3,570; 1,695	2	4.4; 10.1	1; 2.4	nd; 712	2	33; 42	8; 10	14.6; 11.1	2	3.8; 5.2	1	nd	2	9.4(8.03 ± 0.11)	2	100	3

Antagonist concentrations were chosen based on their antagonistic activity on Y1R.

EC50 values were determined after 1 h stimulation by increasing concentration of NPY or NPY together with different antagonists. Data are shown as means from at least three independent experiments (n>2) or results of two individual experiments (n=2) each performed in technical duplicate.

Sample size n, the number of independent experiments performed in technical duplicate.

The EC50 ratio represents the shift between the NPY and NPY/antagonist curve (EC50(NPY+antagonist) / EC50(NPY)) and characterizes the antagonistic effect on wild-type receptor or receptor mutants. By comparison of EC50 ratios between wild type and receptor mutant, influences of all tested residues on antagonist activity were determined. The higher ratio, the higher antagonist activity. A reduced EC50 ratio of mutant compared to the wild-type receptor was interpreted as important for the respective antagonist.

Kb values were determined using the Gaddum transformation (Kb = [Antagonist] / (EC50 ratio – 1)).

These data were obtained at a reduced concentration of UR-HU404 (10−8 M) as concentration response curves did not reach saturation (EC50 > 10,000 nM) when high concentration was used (10−7 M).

nd: not determined; /: not tested.

The hydroxyphenyl group of n class="Chemical">UR-MK299 sits in a groove formed by helices III and VI of the receptor, enabling hydrophobic contacts with residues Q1203.32, C1213.33, I1243.36, W2766.48 and L2796.51. In Y1R and Y2R, Q1203.32 is suggested to be the interaction partner for the C terminus of NPY and crucial for receptor activation[16]. In the Y1R–UR-MK299 structure, this residue forms a hydrophobic contact with the phenyl ring of the hydroxyphenyl group in UR-MK299, potentially blocking the binding of Y1R to NPY. Mutagenesis data show that the mutation Q1203.32N does not influence the inhibitory effect of Y1R antagonists on NPY signalling, but a replacement to histidine dramatically increases antagonistic activity of these ligands (Fig. 3d, e and Extended Data Table 2), suggesting that an additional π-stacking interaction with the antagonist is beneficial at this position. The highly conserved residue W6.48 represents the so-called “toggle switch” and was suggested to trigger receptor activation through a conformational change in various GPCRs[17]. In the Y1R–UR-MK299 structure, the residue W2766.48 is in a conformation similar to those observed in other inactive class A GPCR structures and distinct from their active-state conformations[18,19]. The hydroxyphenyl group of UR-MK299 forms a hydrophobic contact with W2766.48, potentially preventing its activation-related motion and stabilizing the receptor in an inactive conformation. Compared to the wild-type receptor, the Y1R mutant W2766.48A displayed an over 2,000-fold decrease in binding affinity to [3H]-UR-MK299 (Extended Data Table 3) and reduced the antagonistic activity of the argininamide-type Y1R antagonists by 4-7 fold (Fig. 3f and Extended Data Table 2), supporting its important role in antagonist recognition.

Extended Data Table 3

Binding of Y1R antagonists and agonists to membrane preparations from Sf9 cells expressing wild-type and mutant Y1Rs

a. Binding of antagonists to wild-type and mutant Y1Rs
Y1R mutants	Kd (nM)*		Ki (nM)†
	[3H]-UR-MK299	n‡	BMS-193885	n	BIBP3226	n	BIBO3304	n	UR-MK136	n	UR-MK289	n
Wild type	0.17 ± 0.03	3	22 ± 6	3	2.4; 3.1	2	1.6 ± 0.3	3	2.8; 4.0	2	25; 28	2
Crystallization construct	0.33 ± 0.06	3	38 ± 2	4	/		/		/		/
C1213.33A	1.4; 2.4	2	/		/		/		/		/
I1243.36A	7.0; 8.0	2	9,500 ± 1,700	3	15 ± 5	3	11 ± 2	3	12 ± 2	3	80 ± 16	3
I1243.36F	1.3; 1.9	2	/		/		/		/		/
F1734.60A	9.1 ± 2.2	4	590 ± 220	3	68; 84	2	120 ± 17	4	88; 110	2	660 ± 110§	5
F1734.60W	0.31; 0.32	2	110; 130	2	15; 26	2	13 ± 4	3	3.6; 4.3	2	6.8; 9.2	2
T2125.39A	0.12; 0.18	2	150; 150	2	13; 12	2	2.5 ± 0.3	3	5.7; 7.5	2	18; 21	2
L2155.42G	4.2; 5.6	2	9.6 ± 1.9	4	43; 23	2	47 ± 4	3	11; 8.4	2	29; 30	2
Q2195.46A	4.1; 5.1	2	0.50 ± 0.07	4	35; 53	2	6.2 ± 0.7	3	16 ± 4	3	13 ± 4	3
Q2195.46V	5.1 ± 1.2	3	/		/		/		/		/
W2766.48A	>500	3	/		/		/		/		/
L2796.51A	1.0; 1.1	2	160; 220	2	110; 110	2	13 ± 2	3	120; 150	2	320 ± 40	3
T2806.52A	0.16 ± 0.04	3	7,300 ± 1,300	3	32 ± 7	3	2.6 ± 0.4	3	4.2 ± 1.1	3	8.2 ± 4.2	3
N2836.55A	>500; >500	2	/		/		/		/		/
D2876.59A	>500; >500	2	/		/		/		/		/
F3027.35A	>500; >500	2	/		/		/		/		/

Dissociation constant determined by saturation binding at Sf9 membranes (receptor expression was confirmed by western blot analysis) using a sodium-containing buffer (a) or a sodium-free buffer (b) (Note: the sodium-free buffer was used for the determination of agonist binding affinity because porcine NPY exhibited approx. 10-fold higher affinity in the sodium-free buffer compared to the sodium-containing buffer (data not shown)).

Dissociation constant determined by competition binding with [3H]-UR-MK299 at Sf9 membranes using a sodium-containing buffer (a) or a sodium-free buffer (b).

Sample size, the number of independent experiments performed in technical triplicate. If n>2, data are shown as means ± s.e.m. If n=2, results of two individual experiments are shown.

The lower curve plateau of the four-parameter logistic fit, amounting to 17±3% of specifically bound [3H]-UR-MK299 (mean ± s.e.m. from five independent experiments), was different from zero (P < 0.005, one-sample one-tailed t-test), which is indicative of a non-competitive mechanism.

/: not tested.

Residues N2836.55 and n class="Chemical">D2876.59 were suggested as the most important amino acids for Y1R ligand recognition[20]. In the Y1R–UR-MK299 structure, N2836.55 is engaged in two hydrogen bonds with the alpha-nitrogen and the carboxylic oxygen of arginine in UR-MK299. D2876.59 builds a salt bridge with the protonated guanidinyl moiety and a hydrogen bond with the carbamoyl group, in agreement with a decrease in antagonist affinity when the carbamoyl group was replaced by an alkoxycarbonyl, acyl or alkyl group[21]. The mutants N2836.55A and D2876.59N displayed a dramatic loss of NPY-induced receptor function and a complete abolishment of antagonistic activity for the small-molecule antagonists (Fig. 3g, h and Extended Data Table 2), which go along with an over 2,000-fold decreased binding affinity of Y1R to [3H]-UR-MK299 (Extended Data Table 3). These data indicate the critical roles of these two residues in agonist- and antagonist-binding. Additionally, the carbamoyl substituent at the guanidine group binds deep in a sub-pocket shaped by helices V and VI, characterized by hydrophobic contacts with L2165.43, T2806.52 and N2836.55, and a hydrogen bond between the oxygen of the propionyl group and Q2195.46. The latter was reflected by a 30-fold decrease in binding affinity of [3H]-UR-MK299 to the Y1R mutants Q2195.46A and Q2195.46V (Extended Data Table 3). Extra empty space at the bottom of the sub-pocket is observed in the Y1R–UR-MK299 structure, suggesting that a larger substituent may be allowed (Extended Data Fig. 1c). This is supported by studies showing that some other carbamoylated argininamide-type Y1R antagonists containing longer carbamoyl chains, such as UR-MK136 (Extended Data Fig. 1j), bind to the receptor with a relatively high affinity[10].

UR-MK299 was reported to exhibit high Y1R selectivity (Y2R: Ki>3,000 nM, Y4R/n class="Chemical">Y5R: Ki>10,000 nM) and specificity compared to two related neuropeptide FF (NPFF) receptors (NPFF1R: Ki=1,000 nM; NPFF2R: Ki>3,000 nM)[10]. Sequence alignment reveals that most of the key residues involved in UR-MK299 binding are conserved between Y1R, the other NPY receptors and the NPFFRs, except for F4.60, Q5.46, N6.55 and F6.58 (Extended Data Fig. 3), indicating the importance of these four residues in terms of selectivity and specificity of UR-MK299. Y2R is the only NPY receptor with L5.46 instead of Q5.46, preventing key polar contacts. In Y4R, E6.58 disturbs the F6.54/F6.58/F7.35 hydrophobic patch and most likely mediates selectivity, supported by the F6.58E mutation in Y1R dramatically reducing binding affinity for BIBP3226[22], which contains the same diphenylmethyl group as UR-MK299. Similarly, hydrophilic residues at key positions impede high-affinity binding at Y5R (T6.58) and NPFF2R (S6.58), while the hydrophobic pocket is preserved though with less bulk in NPFF1R (L6.54/I6.58/F7.35), leading to moderate affinity of BIBP3226 (Ki=18 nM)[10].

Extended Data Figure 3

Sequence alignment of the human NPY receptors and the human NPFF receptors

Colours represent similarities of residues: red background, identical; red, strongly similar. Key residues in the UR-MK299 binding pocket, which are conserved or variable among receptors, are indicated by red and black arrows, respectively. The alignment was generated by uniprot (http://www.uniprot.org/align/) and graphic was prepared on ESPript 3.0 server (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi).

The ligand BMS-193885 occupies a similar binding pocket as UR-MK299 within the helical bundle of Y1R (Fig. 2c, d and Extended Dan class="Chemical">ta Fig. 1k). The dihydropyridine group fits in a sub-pocket formed by helices III, V and VI, which aligns with previous structure-activity relationship (SAR) studies showing that larger substituents at the position 3 of the dihydropyridine ring dramatically reduced Y1R binding affinity[23]. Residue T2806.52 forms a hydrogen bond with the nitrogen of the dihydropyridine ring as confirmed by our mutagenesis studies showing that the mutation T2806.52A decreased the binding affinity of BMS-193885 by about 330 fold (Extended Data Table 3) and reported for a lower affinity N-methylated derivative[24]. Additionally, the dihydropyridine ring makes a hydrophobic contact with residue I1243.36, which is consistent with a 400-fold decrease in affinity at the mutant I1243.36A (Extended Data Table 3). It was also reported that methylation of either nitrogen of the urea group of BMS-193885 significantly decreased binding ability of the methylated derivatives to Y1R[24], suggesting that these hydrogen-bond donors are critical for Y1R recognition. Indeed, in the BMS-193885-bound Y1R structure, the urea group forms hydrogen bond interactions with D2876.59. Similar to the diphenylmethyl group of UR-MK299, the piperidine and methoxyphenyl rings of BMS-193885 form extensive hydrophobic contacts with the residues F2826.54, F2866.58 and F3027.35. Replacement of the methoxyphenyl substituent by piperidine resulted in lower binding affinity to Y1R[23], indicating the importance of the methoxyphenyl group in Y1R binding and reflecting lipophilic demands at this position.

Understanding the binding mode of the endogenous agonist n class="Gene">NPY at a molecular level will facilitate the rational development of Y1R selective ligands. The C-terminal pentapeptide of NPY was identified to be essential for binding to the NPY receptors[25]. Since the hydroxyphenyl and the argininamide group of UR-MK299 mimic R35 and Y36 in the C terminus of NPY (Extended Data Fig. 1l), the crystal structure of Y1R–UR-MK299 serves as a good template for molecular docking of the agonist. To aid docking, complementary mutagenesis studies were performed to determine corresponding interaction partners between Y1R and NPY (Extended Data Table 4a). Furthermore, solid-state NMR chemical shift measurements revealed residue-specific alterations of the secondary structure of NPY upon binding to Y1R (Extended Data Fig. 4). A number of key Y1R/NPY contacts identified by the mutagenesis studies were used to guide NPY docking in Rosetta[26] with the final models being filtered against the NMR data to generate a final ensemble that best represents the combined data. The NPY-bound model reveals a relatively flat binding pose of NPY to Y1R with the C-terminal tetrapeptide R33-Y36, identified as a random coil/β-strand structure in NMR, penetrating into the binding pocket (Fig. 4a). The unstructured N terminus, Y1-P13, is in close proximity to the second extracellular loop (ECL2), while the α-helix in the middle region of NPY, A14-T32, lays along ECL1 and ECL3 and points away from ECL2.

Extended Data Table 4

IP accumulation assays of wild-type (WT) and mutant Y1Rs for NPY/NPY analogues

a. IP accumulation assays of complementary mutagenesis between NPY/NPY analogues and WT and mutant Y1Rs
Peptides*	WT			Q 1203.32H			I293ECL3N			N299732A			N2836.55A			D2876.59A‖
	EC50 (nM)† (pEC50 ± SEM)	X-fold over WT‡	n§	EC50 (nM) (PEC50 ± SEM)	X-fold over WT	n	EC50 (nM) (PEC50 ± SEM)	X-fold over WT	n	EC50 (nM) (PEC50 ± SEM)	X-fold over WT	n	EC50 (nM) (PEC50 ± SEM)	X-fold over WT	n	EC50 (nM) (PEC50 ± SEM)	X-fold over WT	n
NPY	1.5 (8.83 ± 0.02)	1	51	39 (7.41 ±0.11)	26	3	169 (6.77 ± 0.08)	113	8	91 (7.04 ±0.09)	61	5	1,053 (5.98 ±0.13)	702	6	1,384 (5.86 ±0.09)	461	3
[N30]NPY	289 (6.54 ± 0.06)	1	9	1	/	/	3116; 1274	11; 4	2	/	/	/	/	/	/	/	/	/
[A33]NPY	5,395 (5.27 ± 0.07)	1	3	/	/	/	/	/	/	5,306 (5.28 ± 0.07)	1	3	/	/	/	/	/	/
[A35]NPY	>10,000	1	3	/	/	/	/	/	/	/	/	/	/	/	/	>4,900	>5	3
[A36]NPY	1,378 (5.86 ± 0.06)	1	6	/	/	/	/	/	i	/	/	/	nd	nd	3	/	/	/
NPY-tyramide	68 (7.17 ±0.11)	1	9	nd	nd	3	/	i	i	/	/	/	/	i	/	/	/	/

Peptides were synthesized following the methods described in peptide synthesis section of Methods.

EC50 were determined using GraphPad Prism 5.0. All curves were normalized to the top and bottom values of the Y1R/NPY curve. Nonlinear regression (curve fit) was performed for normalized response in all assays. All data are shown as means from at least three independent experiments (n>2) or results of two individual experiments (n=2) each performed in technical duplicate.

The EC50 shifts were determined by EC50(mutant) / EC50(WT), set Hill slope to 1. For the wild-type receptor x-fold is set to 1. Lower EC50 shift of NPY analogue/mutant compared to NPY/mutant was interpreted as no further loss of function and a direct interaction between both positions.

Sample size n, the number of independent experiments performed in technical duplicate.

Data are from reference [27].

nd, not determined up to 10−4 M agonist concentration; /, not tested.

Extended Data Figure 4

Pharmacological characterization of refolded Y1R and NMR studies of Y1R-bound NPY

a, Binding of atto520-labelled NPY (50 nM) to increasing amounts of Y1R-containing bicelles or empty bicelles. Data reflect fluorescence enhancement upon binding. An inflection point at EC50 = 52 nM was determined. Two independent experiments were performed in technical duplicate with similar results. Data shown are from a representative experiment. b, Typical 13C MAS single quantum (SQ) / double quantum (DQ) correlation spectrum of NPY in the presence of Y1R reconstituted into large bicelles at −30 °C. NMR spectra were acquired from 1 to 3 independent preparations for each labelled amino acid with similar results (see panel d). Data shown are from a representative experiment. c, Table showing 13C NMR chemical shifts of assigned amino acids of NPY bound to Y1R (referenced to tetramethylsilane) as acquired in solid-state NMR experiments. d, 13C chemical-shift index of NPY bound to Y1R in large DMPC/DHPC-c7 bicelles (q > 20) compared with docked models. Plotted is the measured chemical shift difference (Cα-Cβ) for each individual residue of NPY subtracted by chemical shift difference of the same amino acid type in random coil conformation in black. Individual data points from 1 to 3 independent experiments for each labelled amino acid are shown. Typical experimental errors in determining chemical shifts under these conditions are ±1 ppm. Chemical shifts were back-calculated for the top docking solutions and filtered against the experimental data to generate a final ensemble of docked poses. Their average chemical-shift index and associated standard deviation from 10 top docked poses are shown in red.

Figure 4

Docking poses of NPY

a, Predicted NPY binding pose. The receptor and the lowest energy NPY conformation are shown as cartoons, and coloured brown and cyan, respectively. b, Comparison of UR-MK299 (yellow sticks) binding mode and predicted ensemble binding mode of NPY residues R35 and Y36 (cyan sticks). c. Predicted binding mode between Y1R and NPY C terminus. Key Residues involved in Y1R-NPY interaction are shown as sticks and coloured dark brown (Y1R) and blue (NPY). d, Predicted binding mode between Y1R’s ECL2 and NPY N terminus. Residues in Y1R’s ECL2 and NPY N terminus that may form contacts are shown as dark brown and blue sticks, respectively.

Inspection of the NPY binding pocket shows a similar binding pose of n class="Gene">NPY’s residue R35 as the argininamide of UR-MK299 (Fig. 4b). R35 forms a salt bridge with the residue D2876.59 of Y1R and approaches to N2836.55 (Fig. 4c). The NPY mutant R35A displays an over 6,000-fold decrease in activity and represents the highest influence on agonist potency of all tested NPY analogues (Extended Data Table 4a), supporting the importance of the positively charged residue in NPY recognition. D/E is not found at position 6.59 in any peptide GPCRs except for the receptors that bind to RF-amide peptides, including NPFF, prolactin releasing peptide and pyroglutamylated Arg-Phe-amide peptide, which share a common C-terminal Arg-Phe-NH2 motif, supporting the hypothesis that the arginine may function similarly as R35 of NPY by interacting with the conserved D/E6.59 of the respective receptors[27]. In contrast to the similar binding mode between R35 of NPY and the guanidine group of UR-MK299, the C-terminal tyrosinamide of NPY and the hydroxyphenyl group of UR-MK299 show different orientations. The hydroxyphenyl ring is oriented towards helix V (Q2195.46) in the UR-MK299-bound Y1R structure, while Y36 of NPY points toward the residue Q1203.32 on helix III in the NPY-docked model (Fig. 4c). This may arise from the opposite configuration of the stereo center in R35 of NPY and UR-MK299, as well as by only partial mimicking of the Y36-NH2 of NPY by a 4-hydroxybenzyl group in UR-MK299 (Extended Data Fig. 1g, l). In the Y1R–UR-MK299 structure, residue Q1203.32 forms a hydrophobic contact with the hydroxyphenyl group of the antagonist. In contrast, the NPY-bound model shows that side chain of Q1203.32 points almost in the opposite direction and engages in a hydrogen bond with the hydroxy group of Y36-NH2 (Extended Data Fig. 1d), similar as the interaction between the Y2R residue Q3.32 and NPY suggested by Xu et al.[28] In Y2R, it was also reported that Q3.32 may interact with the C-terminal amide of NPY[16]. IP accumulation studies show that the Y1R mutation Q1203.32H leads to a 26-fold decrease in potency of NPY, and NPY-tyramide lacking the C-terminal amide displays a 45-fold loss of activity. Complementary mutagenesis revealed an additional reduction of NPY-tyramide potency at the mutant Q1203.32H, and thus rules out a direct contact between the C-terminal amide of NPY and Q1203.32 in Y1R (Extended Data Table 4a). Additionally, Y36 of NPY forms hydrophobic contacts with Y1002.64 and W1062.70 in Y1R (Fig. 4c). Although Y1002.64 is not involved in antagonist binding, mutagenesis data suggest a critical role in agonist recognition as the Y1R mutant Y1002.64A displays a 284-fold decrease in potency for NPY (Fig. 3i and Extended Data Table 2). Furthermore, the model reveals close contacts between L30 of NPY and I293 in ECL3 of Y1R and between R33 of NPY and the Y1R residue N2997.32 (Fig. 4c), which align with complemetary mutagenesis data showing no further loss of function for combining mutant I293N with [N30]NPY and N2997.32A with [A33]NPY (Fig. 3j-m and Extended Data Table 4a).

Previous studies showed different behaviors of the NPY receptors binding to the N terminus of n class="Gene">NPY. While Y2R and Y5R can bind to N-terminally truncated NPY, Y1R and Y4R require the complete N terminus of NPY for full agonist potency[25,27]. However, these data did not allow conclusions about the interaction of the N terminus of NPY with the receptor. To further explore an involvement of NPY’s N terminus in receptor-ligand recognition, we performed mutagenesis studies, showing that truncation of the first two residues of NPY (NPY(3-36)) reduces peptide potency by more than 50 fold (Extended Data Table 4b). This decrease in potency, however, is not seen when these residues are mutated to alanine ([A1,A2]NPY: 5-fold EC50 shift), suggesting important contributions of peptide backbone in binding to the receptor. Our NPY-bound model suggests that the N-terminal region of NPY makes close contacts with the fragment T180-F199 in ECL2 of Y1R and is also in proximity to the receptor N terminus (Fig. 4a, d). To experimentally verify interacting sites in the receptor, we performed photo-crosslinking studies between NPY analogues carrying the highly reactive p-benzoyl-phenylalanine [Bpa1, K4[(Ahx)2-biotin]]NPY and Y1R. Crosslinked fragments were assigned to two regions in Y1R, the N terminus (K21-D32) and ECL2 (A191-D205) (Extended Data Fig. 5 and Extended Data Table 5). Previous studies demonstrated that deletion of the Y1R N terminus does not interfere with receptor signalling, but reduces NPY binding by about 95% compared to the full-length receptor[29]. This creates the possibility that the N terminus of Y1R plays a role in recognizing and positioning the peptide ligand, which aligns with the photo-crosslinking data. Consistent with the crosslinking hits in receptor ECL2, our mutagenesis data show that mutations F184A/N and V197N in this region greatly reduce NPY potency (Extended Data Table 4c). Together, these data suggest that the N terminus and ECL2 of Y1R play important roles in recognition of NPY’s N terminus and receptor activation. This contrasts with NPY binding at Y2R, in which ECL2 may interact with the central α-helix of NPY and the peptide N terminus is flexible and unanchored by the receptor[16]. Whereas this study provides insights into the interactions between Y1R and NPY, further structural details, such as the structures of Y receptors bound to NPY, are required to fully understand the endogenous agonist binding modes of the NPY receptor family.

Extended Data Figure 5

Photo-crosslinking experiments between NPY and Y1R

a, Mass spectra of photo-crosslinked Y1R with [Bpa1,K4[(Ahx)2-biotin]]NPY. Exemplary MALDI-ToF mass spectra of photo-crosslinked samples enzymatically digested by rLys-C and Glu-C. Potential Y1R fragments are labelled. Two independent experiments were performed with similar results. N, N terminus of Y1R (blue); E, ECL2 (red). b, Respective regions of NPY N terminus at Y1R. Amino acid sequence of Y1R with a C-terminal His-tag. The two detected regions within Y1R (N terminus (blue), ECL2 (red)) after crosslinking with [Bpa1,K4[(Ahx)2-biotin]]NPY are emphasized in boxes. The different sizes of the boxes represent different detected fragments (Extended Data Table 5), respectively. Experiments were repeated twice independently with similar results, and only fragements that were observed in both experiments are listed here and in Extended Data Table 5. c, Binding of atto520-labelled NPY (50 nM) to increasing amounts of cell-free produced Y1R in Brij-58. Data reflect fluorescence enhancement upon binding. EC50 value of 69 nM was determined. Data shown are means ± s.e.m. from six independent experiments performed in technical triplicate.

Extended Data Table 5

Mass spectromeric signals and calculated mass of photo-crosslinked Y1R with [Bpa1,K4[(Ahx)2-biotin]]NPY

MALDIToF MS (m/z)*	Number in Y1R	Position [Bpa1,K4[(Ahx)2-biotin]]NPY	Mcalc.(Da)†	[Mcalc + (Da)	[Mcalc + Na]+(Da)	[Mcalc + K]+ (Da)
1824.2			not identified
1867.3			not identified
1892.6	11–32	1 – 6	3760.7	3761.8	3783.7	3799.7
1988.2	–	1–11 + 5–6	1986.9	1987.9	2009.9	2025.9
2001.1	–	1–10 + 5–7	2000.0	2001.0	2022.9	2038.9
2059.2	–	1–10 + 8–11	2057.9	2058.9	2080.9	2096.9
	–	1–11 +8–10	2057.9	2058.9	2080.9	2096.9
2073.2			not identified
	–	1–10 + 8–11	2057.9	2058.9	2080.9	2096.9
2081.3	–	1–11 +8–10	2057.9	2058.9	2080.9	2096.9
2121.3			not identified
2162.2	191 – 194	1 – 10	2139.0	2140.0	2162.0	2178.0
2311.4			not identified
2317.4			not identified
2381.4	191 – 195	1–11	2380.1	2381.1	2403.1	2419.1
2381.8	195–205	1 – 4	2381.1	2382.1	2404.1	2420.1
2397.3	32–32	1 – 16	2374.0	2375.0	2397.0	2413.0
	194–194	1 – 16	2374.0	2375.0	2397.0	2413.0
	21 – 21	1 – 16	2387.1	2388.1	2410.1	2426.1
2410.3	195 – 195	1 – 16	2387.1	2388.1	2410.1	2426.1
	194–195	1 – 15	2387.1	2388.1	2410.1	2426.1
2413.3	32–32	1 – 16	2374.0	2375.0	2397.0	2413.0
	194 – 194	1 – 16	2374.0	2375.0	2397.0	2413.0
2450.4	–	1–15 + 8–10	2427.1	2428.1	2450.1	2466.1
2514.5			not identified
2553.5	194 – 200	1 – 10	2530.1	2531.1	2553.1	2569.1
2589.6			not identified
2775.7			not identified
2807.7			not identified
3031.7	22–29	1 – 15	3030.4	3031.4	3053.4	3069.4
3377.7	191 – 200	1 – 15	3376.5	3377.5	3399.5	3415.5
3784.1	11–32	1 – 6	3760.7	3761.8	3783.7	3799.7
3966.3	–	1–7 + 17–36	3943.1	3944.1	3966.1	3982.1
4524.9			not identified

Determined signals by MALDI-ToF MS.

Selected calculated masses in Dalton of possible photo-crosslinked fragments of Y1R with [Bpa1,K4[(Ahx)2-biotin]]NPY or [Bpa1,K4[(Ahx)2-biotin]]NPY with itself. The fragments are selected based on the correlation with the detected signals. For clarity, further calculated masses of possible photo-crosslinked fragments are not shown.

Methods

Cloning and protein expression

DNA sequence of wild-type human Y1R was optimized and synthesized by Genewiz and then cloned into a modified pFastBac1 vector (Invitrogen), which conn class="Chemical">tains an expression cassette with a haemagglutinin (HA) signal sequence followed by a Flag tag prior to the receptor at the N terminus and a PreScission protease site followed by a 10×His-tag at the C terminus. An engineered construct was generated by inserting a modified T4 Lysozyme (T4L)[31] at the third intracellular loop (ICL3) between residues R241 and D250 and introducing a mutation F1293.41W[32]. Twenty-five amino acids (V359-I384) were truncated at the C terminus to further improve protein yield and stability. Bac-to-Bac Baculovirus Expression System (Invitrogen) was used to generate high-titer (>108 viral particles per ml) recombinant baculovirus. Spodoptera frugiperda (Sf9) cells (Invitrogen) at density of 2 × 106 cells per ml were infected by viral stock at MOI (multiplicity of infection) of 5. In company with the virus, a ligand (UR-MK299 or BMS-193885) was added to the cell culture to a final concentration of 1 μM. Transfected cells were cultured at 27 ºC for 48 h and then collected by centrifugation and stored at −80 ºC until use.

Purification of Y1R–UR-MK299 and Y1R–BMS-193885 complexes

Frozen insect cells expressing the Y1R–UR-MK299 complex were disrupted with thawing and repeated dounce homogenization in a n class="Disease">hypotonic buffer containing 10 mM HEPES, pH 7.5, 10 mM MgCl2, 20 mM KCl and protease inhibitor cocktail (Roche). After centrifugation at 160,000 g for 30 min, cell debris was re-suspended in a high osmotic buffer (10 mM HEPES, pH 7.5, 1 M NaCl, 10 mM MgCl2, 20 mM KCl) followed by extensive homogenization. Soluble and membrane associated proteins were removed from the suspension by centrifugation. This procedure was repeated for 2-3 more times and then the hypotonic buffer was used to remove the high concentration of NaCl. Purified membranes were re-suspended in the hypotonic buffer with additional 30% (v/v) glycerol and stored at −80 ºC until use.

Purified membranes were thawed on ice in the presence of 100 μM n class="Chemical">UR-MK299, 2 mg ml−1 iodoacetamide (Sigma) and EDTA-free protease inhibitor cocktail (Roche) and incubated at 4 ºC for 1 h. Equal volume of solubulization buffer containing 100 mM HEPES, pH 7.5, 1 M NaCl, 1% (w/v) n-dodecyl-b-D-maltopyranoside (DDM, Anatrace), 0.2% (w/v) cholesterol hemisuccinate (CHS) (Sigma) was added and incubation was continued for additional 3 h. The supernatant was isolated by centrifugation at 160,000 g for 30 min and incubated with TALON resin (Clontech) supplemented with 10 mM imidazole, pH 7.5 at 4 ºC overnight. The resin was then washed with ten column volumes of 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol, 30 mM imidazole and 50 μM UR-MK299, followed by ten column volumes of 50 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol, 10 mM MgCl2, 5 mM ATP and 50 μM UR-MK299 and five column volumes of 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol and 50 μM UR-MK299. The protein was eluted by five column volumes of 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol, 300 mM imidazole and 100 μM UR-MK299. A PD MiniTrap G-25 column (GE healthcare) was used to remove imidazole. The C-terminal His-tag and glycosylation was then treated by overnight digestion with His-tagged PreScission protease (custom-made) and His-tagged PNGase F (custom-made). Ni-NTA super flow resin (Qiagen) reverse binding was performed to remove the PreScission protease, PNGase F and the cleaved His-tag. The purified Y1R–UR-MK299 complex was collected and concentrated to 20-30 mg ml−1 with a 100 kDa molecular weight cutoff concentrator (Sartorius Stedim Biotech). Receptor purity and monodispersity were estimated by SDS-PAGE and analytical size-exclusion chromatography (aSEC).

The Y1R–BMS-193885 complex protein was purified following the same procedn class="Chemical">ure as above. The membranes of Y1R construct were incubated with 50 μM BMS-193885, 2 mg ml−1 iodoacetamide (Sigma), and EDTA-free protease inhibitor cocktail (Roche) at 4 ºC for 1 h, and then solubilized in final concentration of 50 mM HEPES, pH 7.5, 500 mM NaCl, 0.5% (w/v) DDM, 0.1% (w/v) CHS, 10% glycerol and 25 μM BMS-193885 at 4 ºC for 3 h. The solubilized Y1R–BMS-193885 complex bound to the TALON resin was washed with ten column volumes of 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol, 30 mM imidazole and 25 μM BMS-193885, followed by ten column volumes of 50 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol, 10 mM MgCl2, 5 mM ATP and 25 μM BMS-193885 and five column volumes of 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol and 25 μM BMS-193885. The protein was eluted by five column volumes of 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% (v/v) glycerol, 300 mM imidazole and 50 μM BMS-193885. The eluted sample was concentrated and desalted using the PD MiniTrap G-25 column (GE healthcare). Overnight digestion by Precission protease and PNGase F and Ni-NTA reverse binding were then performed to further purify the protein. The complex protein was concentrated to 10-20 mg ml−1 and analyzed by SDS-PAGE and aSEC.

Lipidic cubic phase crystallization of antagonist-bound Y1Rs

The Y1R sample in complex with UR-MK299 or BMS-193885 was mixed with molten n class="Chemical">lipid (monoolein/cholesterol 10:1 by mass) at a weight ratio of 1:1.5 (protein:lipid) using two syringes to create lipidic cubic phase (LCP). The mixture was dispensed onto glass sandwich plates (Shanghai FAstal BioTech) in 40 nl drop and overlaid with 800 nl precipitant solution using a Gryphon robot (Art-Robbins). Protein reconstitution in LCP and crystallization trials were performed at room temperature (19-22 ºC). Plates were placed in an incubator (Rock Imager, Formulatrix) and imaged at 20 ºC automatically following a schedule. Crystals of Y1R–UR-MK299 complex showed up after 4 days and grew to the full size (150 × 50 × 5 μm3) within two weeks in 0.1 M Tris, pH 7.4-8.0, 30-40% (v/v) PEG400, 50-150 mM sodium tartrate and 100 μM UR-MK299. The Y1R–BMS-193885 complex was crystallized in 0.1 M HEPES, pH 7.2-7.6, 20% PEG400 and 50 μM BMS-193885 with the maximum size of 30 × 10 × 5 μm3. The crystals of Y1R–UR-MK299 and Y1R–BMS-193885 complexes were harvested directly from LCP using 150 μm and 50 μm micro mounts (M2-L19-50/100, MiTeGen), respectively, and flash frozen in liquid nitrogen.

Data collection and structure determination

X-ray diffraction data were collected at the SPring-8 beam line 41XU, Hyogo, Japan, on a Pilatus3 6M detector (X-ray wavelength 1.0000 Å). Crysn class="Chemical">tals were exposed with a 10 μm × 8 μm mini-beam for 0.2 s and 0.2º oscillation per frame. Data from 47 best-diffracting crystals of the Y1R–UR-MK299 complex and 33 crystals of the Y1R–BMS-193885 complex were processed by XDS[33], respectively. Structure of the Y1R–UR-MK299 complex was solved by molecular replacement (MR) implemented in Phaser[34] using the receptor portion of NTSR1 (PDB ID: 4GRV), converted to polyalanines, and T4L structure (PDB ID: 1C6P) as search models. The correct MR solution contained one Y1R-T4L molecule in the asymmetric unit. Initial refinement was performed with REFMAC5[35] and BUSTER[36], and then manual examination and rebuilding of the refined coordinates were carried out in COOT[37] using both |2Fo| – |Fc| and |Fo| – |Fc| maps. The structure has been carefully refined and the ramachandran plot analysis indicates that 100% of the residues are in favorable (95.5%) or allowed (4.5%) regions (no outliers). The final model includes 312 residues (F18-R241 and D250-F337) of the 384 residues of Y1R and residues N2-Y161 of T4L. The Y1R–BMS-193885 complex structure was solved using Y1R in the Y1R–UR-MK299 complex and T4L as search models and refined under the same procedure. The ramachandran plot analysis indicates that 100% of the residues are in favorable (95.4%) or allowed (4.6%) regions (no outliers). The final model of the Y1R–BMS-193885 complex contains 301 residues (D31-R241 and D250-D339) of Y1R and the 160 residues of T4L. Helix VIII in the Y1R–UR-MK299 structure rotates towards helix VI by about 90° compared to that in the BMS-193885-bound structure most likely due to crystal packing (Extended Data Fig. 1).

Immunoblotting

Total solublized protein of Sf9 membrane preparations (see above) used in radio ligand binding assay was determined by the Bradford method according to the manufactn class="Chemical">urers’ protocol (BioRad Protein Assay; BioRad, Munich, Germany). Aliquots of homogenized membrane preparations, corresponding to 100 μg of protein, were centrifuged at 50,000 g at 4 °C for 15 min, and the pellets were re-suspended in 50 mM Tris, pH 7.4, supplemented with 1 mM EDTA and protease inhibitors (SIGMAFAST Protease Inhibitor cocktail tablets, Sigma) at a protein concentration of 1,600 μg ml−1. Membrane homogenates (15 μl) were processed and subjected to immunoblotting as described previously[38] with the following modifications: blotting onto the nitrocellulose membrane was performed at 60 mA for 60 min. Primary antibody ANTI-FLAG M1 from mouse (Sigma, order no. F3040, lot SLBK1592V) was diluted 1:500. The secondary antibody, an anti-mouse IgG HRP-conjugated antibody from goat (Sigma, order no. A0168, lot 080M4839) was diluted 1:80,000. The washing steps after incubation with the primary and the secondary antibody were 3 × 10 min each. Control experiments in the absence of the primary antibody were not performed.

Radio ligand binding assay

All binding experiments with [3H]-n class="Chemical">UR-MK299 (synthesis described elsewhere[10]) were performed at Sf9 membrane preparations in PP 96-well microplates (Greiner bio-one) at 23±1 °C using a sodium-containing, iso-osmotic HEPES buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM KCl, 2.5 mM CaCl2, 1.2 mM KH2PO4, 1.2 mM Mg2SO4 and 25 mM NaHCO3 supplemented with 1% BSA) for competition binding studies with antagonists, and a sodium-free, hypo-osmotic HEPES buffer (25 mM HEPES, pH 7.4, 2.5 mM CaCl2 and 1 mM MgCl2 supplemented with 1% BSA) for competition binding studies with the agonist NPY (in the following, both buffers are referred to as ‘binding buffer’). Prior to competition binding experiments, Kd values of [3H]-UR-MK299 were determined by saturation binding using the respective binding buffer. In case of saturation binding experiments, [3H]-UR-MK299 was 1:1 diluted with ‘cold’ UR-MK299 (in the following, the mixture is referred to as ‘radioligand’). On the day of the experiment, Sf9 membranes were thawed and re-suspended using a 1-ml syringe equipped with a needle (20G) followed by centrifugation at 16,000 g at 4 °C for 10 min. The supernatant was discarded and the pellets were re-suspended in binding buffer using a 1-ml syringe equipped with a needle (27G3/4). The membrane homogenates were stored on ice until use. The total amount of protein per well was between 0.25 and 8 μg, depending on the receptor expression level.

Saturation binding experiments: for the determination of ton class="Chemical">tal binding, wells were pre-filled with binding buffer (160 μl), followed by the addition of binding buffer (20 μl), containing the radioligand at a 10-fold higher concentration compared to the final concentration. For the determination of unspecific binding (in the presence of UR-MK299 at a 100-fold excess), wells were pre-loaded with binding buffer (140 μl), binding buffer (20 μl) containing UR-MK299 (10-fold concentrated) and binding buffer (20 μl) containing the radioligand (10-fold concentrated). To all wells, 20 μl of the membrane suspension were added, and the plates were shaken at 23 °C for 90 min. The membranes were collected on GF/C filter mats (0.26 mm; Whatman, Maidstone, UK) (pre-treated with 0.3% polyethylenimine for 30 min) and washed with cold Tris buffer (91 g l−1 Tris base, 25.5 g l−1 MgCl2·6H2O and 3.76 g l−1 EDTA) using a Brandel Harvester (Brandel, Gaithersburg, MD). Filter pieces were punched out and transferred into 1450-401 96-well plates (PerkinElmer, Rodgau, Germany). Rotiscint eco plus (Carl Roth, Karlsruhe, Germany) (200 μl) was added, and the plates were sealed with transparent tape (permanent seal for microplates, PerkinElmer), vigorously shaken for at least 3 h, and kept in the dark for at least 1 h prior to the measurement of radioactivity (dpm) with a MicroBeta2 plate counter (PerkinElmer). Specific binding data (dpm) were plotted against the free radioligand concentration in nM (obtained by subtracting the amount of bound radioligand (nM) (calculated from the specifically bound radioligand in dpm, the specific activity, and the volume per well) from the total radioligand concentration (nM)) and analyzed by a two-parameter equation describing hyperbolic binding (SigmaPlot 11.0, Systat Software Inc., Chicago, IL) to obtain Kd and Bmax values. In case of Kd values < 1 nM, the Bmax was kept below 1,200 dpm by choosing an appropriate protein concentration. In case of Kd values > 1 nM, the Bmax was kept below 3,300 dpm.

Competition binding experiments were performed according to the procedure for satn class="Chemical">uration binding with the following modifications: [3H]-UR-MK299 was used undiluted and in case of Y1R mutants, at which [3H]-UR-MK299 exhibited a Kd value > 3 nM, the total volume per well was 100 μl, i.e., in the case of total binding, wells were pre-filled with binding buffer (80 μl), followed by the addition of binding buffer (10 μl) containing [3H]-UR-MK299 (10-fold concentrated), and the membrane homogenate (10 μl). The following concentrations of [3H]-UR-MK299 were used for competition binding with antagonists: 0.2 nM (wild-type Y1R, T2806.52A, T2125.39A), 0.3 nM (F1734.60W), 1.1 nM (L2796.51A), 5 nM (Q2195.46A), 7 nM (L2155.42G), 10 nM (I1243.36A, F1734.60A). 1 nM [3H]-UR-MK299 was used for competition binding with NPY. The incubation time was 90 min throughout. Unspecific binding was determined in the presence of UR-MK299 (100-fold excess to [3H]-UR-MK299). Total binding was between 700 and 3,500 dpm. Maximum unspecific binding amounted to ≤ 30% of total binding. Specific binding data (dpm) were plotted against log (concentration competitor) and analyzed by a four-parameter logistic equation (log(inhibitor) vs response – variable slope, GraphPad Prism Software 5.0, GraphPad Software, San Diego, CA) to obtain pIC50 values, which were converted to IC50 values. In case of incomplete displacement of [3H]-UR-MK299 (specifically bound radioligand at the highest competitor concentration between 20% and 50%), pIC50 values were determined by plotting log(B/(B0-B)) (Hill plot; B = specifically bound radioligand in the presence of competitor (values between 10 and 90%), B0 = specifically bound radioligand in the absence of competitor (B0 = 100%)) against log(competitor concentration) (at least three data points) and pIC50 values (log(B/(B0-B)) = 0) were determined by linear regression. Ki values were calculated from the IC50 value as well as the respective Kd value (Extended Data Table 3) and the concentration of [3H]-UR-MK299 according to the Cheng-Prusoff equation[39].

IP accumulation assay

The signal transduction assay was performed as previously described[40,41]. Briefly, COS-7 cells (obtained from American Type Culture Collection) were seeded into 48-well plates and were transiently co-transfected with wild-type receptor or receptor mutant and a chimeric G protein (GαΔ6qi4myr) plasmid DNA[42]. Cells were routinely tested for mycoplasma contamination. Cells were radioactively labelled with myo-[2-3H]-Inositol (Perkin Elmer, Boston, USA) overnight, and then stimulated with an increased concentration of NPY (NPY-curve). For antagonist curves, cells were stimulated with the antagonist (BIBP3226: 10−5 M, BIBO3304: 10−6 M, MK-HU404: 10−7 M, UR-MK289: 10−5 M, UR-MK299: 10−7M) parallel to an increased concentration of NPY for 1h (standard conditions). After cell lysis, an anion exchange chromatography was performed and isolated, radioactive accumulated IP derivatives were analyzed by liquid scintillation counting (scintillation cocktail Optiphase HiSafe, Perkin Elmer).

Using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA) the determined concentration response-curves were analyzed. The cn class="Chemical">urves were normalized to the top (100%) and bottom (0%) values of the associated NPY curve. All curves of independent experiments were summarized to one single concentration response-curve by the row means total function. Using nonlinear regression (curve fit) the EC50 and pEC50 ± SEM were examined. The shift between NPY and NPY/antagonist curve is defined as EC50 ratio and calculated by dividing EC50(NPY/antagonist)/EC50(NPY), set Hill slope to 1. All experiments were performed in duplicates of at least two independent experiments.

Live cell fluorescence microscopy

The membrane localization of Y1R and receptor mutants was verified by fluorescence microscopy. n class="CellLine">COS-7 cells were seeded in 8 well μ-slides (IBIDItreat, Martiensried, Germany) and transiently transfected with Lipofectamine2000 transfection reagent (Invivogen, Toulouse, France). 24h post transfection, nuclei were stained with Hoechst33342 (Sigma-Aldrich, Taufkirchen, Germany) and images were recorded using an ApoTome Imaging System, AxioVert Observer Z1 (YFP: Filter Set 46, DAPI: Filter Set 49, ApoTome, 63x/1.40 oil objective, ZEISS, Jena, Germany) in a quasi-confocal setting. The data demonstrate that all the mutants are expressed at similar, and wild-type-like, level in COS-7 cells (Extended Data Fig. 2).

Quantification of receptor surface expression in COS-7 cells

COS-7 cells were seeded into black 96 well plates (Greiner, Kremsmünster, Austria), and transiently transfected with plasmid encoding receptor-eYFP fusion protein using n class="Chemical">MetafectenePro. 24h post transfection, cells were washed once with HBSS, and fluorescence was quantified using a plate reader (Tecan Infinite M200, Tecan, Männedorf, Switzerland) at excitation 488/5 nm and emission 530/5 nm. Data were normalized to mock transfected (0%) and wild type Y1R-eYFP (100%), respectively. Results represent means ± s.e.m. from three independent experiments performed in quadruplicate.

Peptide Synthesis

Porcine NPY (YPSKPDNPGEDAPAEDLARYYSALRHYINLITRQRY-NH2) and n class="Gene">NPY analogues were synthesized by automated solid-phase peptide synthesis (SPPS) on an automated multiple peptide synthesis robot system (Syro, MultiSynTech, Bochum, Germany), using a 9-fluorenylmethoxycarbonyl-tert-butyl (Fmoc/tBu) strategy in 15 μM scale as previously described[43]. NPY-tyramide was synthesized as previously described[44]. Isotopically labelled NPY variants were prepared as described[16], and 13C/15N labelled amino acids were coupled manually with 2 eq. HOBT/DIC in DMF over night. The porcine variant of NPY, which contains a single mutation at position 17 from a methionine to leucine, was used, as this variant has identical binding affinity and signalling properties as human NPY with an increased solubility to assist in handling and will therefore be referred to as wild-type NPY[45].

Modified NPY analogues [Bpa1,K4[(Ahx)2-biotin]]NPY and [K4[(Ahx)2-biotin]]NPY were synthesized by automated solid phase peptide synthesis and Bpa/Ahx/biotin were coupled manually using orthogonal Dde protection groups, cleaved by freshly prepared 3% (v/v) hydrazine in DMF for 10 × 5 min. Manual coupling reactions were performed by incubation of the resin with 5 eq. of the respective amino acid, 5 eq. HOBt and 5 eq. DIC in DMF for 2 h.

For biotin labelling, 3 eq. n class="Chemical">biotin was dissolved in DMF for 10 min at 60 °C. Next, 3 eq. HOBt and 3 eq. DIC were added to the mixture. Coupling was performed overnight at room temperature under constant shaking. Bpa containing peptides were cleaved from the resin and completely deprotected with a mixture of trifluoroacetic acid (TFA)/thioanisole (TA)/water (90:5:5 v/v/v).

All peptides were purified by preparative reversed-phase high-performance liquid chromatography (RP-HPLC) on a Jupiter 4u Proteo RP-C18 column (90 Å, 4 μm, n class="Chemical">Phenomenex), Kinetex 5u XB-C18 column (100 Å, 5 μm, Phenomenex), Kinetex 5u Biphenyl (100 Å, 5 μm, Phenomenex), Aeris 3.6 μm WIDEPORE XB-C18 (200 Å, 3.6 μm, Phenomenex) or Varitide RPC (200 Å, 6 μm, Agilent Technologies). All peptides were characterized by MALDI-ToF (Ultraflex III MALDI-ToF/ToF, Bruker daltonics) and ESI-HCT (High-capacity ion trap ESI-MS, Bruker Daltonics). Peptide purities were determined by two different analytical RP-HPLC systems and using 0.1% (v/v) TFA in H2O (eluant A) and 0.08% (v/v) TFA in acetonitrile (ACN) (eluant B). Purity of all peptides was ≥ 95%.

NMR measurements of Y1R-bound NPY

Fourteen differently isotopically labelled (U-n class="Chemical">13C/15N) porcine NPY peptides were prepared by standard Fmoc solid phase synthesis as described previously[16]. On the basis of the structure of micelle-bound NPY, the positions of the NMR labels were chosen to avoid signal overlap in 13C-13C single quantum/double quantum correlation experiments and to allow straightforward signal assignment.

Expression of the human Y1R in n class="Species">E. coli as inclusion bodies, inclusion body preparation, solubilization of the receptor in SDS and receptor purification was performed according to Schmidt et al.[46] yielding ~40-50 mg Y1R per liter expression medium. To assemble the Y1R into a functional state, a three step folding protocol was applied. In step 1, the purified Y1 receptor was dialyzed against a degassed buffer containing 2 mM SDS, 50 mM NaP, pH 8.5, 1 mM EDTA, 1 mM reduced glutathione, and 0.5 mM oxidized glutathione at room temperature for 48 h. Subsequently, 25% (w/w) poly(ethylene glycol) (M.W. 20 kDa) was added to the buffer to concentrate the receptor before reconstitution. In step 2, bicelles consisting of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) / 1,2-diheptanoyl-sn-glycero-3-phosphocholine (DHPC-c7) (AvantiPolarLipids, Alabaster, USA) with a q value of 0.25 in 50 mM NaP, pH 8.0 were incubated with Y1R, followed by three cycles of fast temperature cycles from 42°C to 0°C. Aggregated protein was removed instantly by centrifugation. In step 3, the Y1R samples were concentrated in large bicelles (q > 20), which were used instead of liposomes because the high achievable receptor packing[46]. In large bicelles, all receptor binding sites are fully accessible. Subsequently, 50 mg ml−1 BioBeadsSM2 was added at least twice to the solution. After removal of the beads with a sieve, samples were washed four times through cycles of pelleting by centrifugation and resolubilization in 50 mM NaP, pH 7.0. Concentration determination of the membrane embedded receptor was performed by solubilization of the bicelles in 10 times the volume of 15 mM SDS and 50 mM NaP, pH 7.0 and subsequent measurement of the Y1R intrinsic absorption at 280 nm using UV-Vis. Labelled NPY variants in a slight molar excess were added to the Y1R after detergent removal but before concentrating.

Assessment of the binding affinity of the Y1R was carried out using homogenous fluorescence assays as described in the literature[47]. The reconstituted receptor was incubated in increasing concentrations with 50 nM fluorescently labelled n class="Gene">NPY (NPY-atto520) overnight at room temperature in 50 mM NaP, pH 7.0 and 0.1% BSA. The fluorescence spectra were recorded on a FluoroMax-2 (JOBIN YVON) in a 10 mm quartz cuvette at 20 °C. The maximum signals of each spectrum were determined, normalized and plotted against the receptor concentration. The inflection point for Y1R binding was determined (OriginPro 8G / DoseResp) at EC50 = 52 nM, demonstrating high functionality of the system. As a control, we used empty bicelles in concentrations, according to the bicelles concentration of the receptor samples, resulting in a lower binding ability to the ligand in comparison to the Y1R.

NMR spectra were recorded on a Bruker Avance III 600 Nn class="Disease">MR spectrometer. The 13C cross-polarization magic angle spinning (CP MAS) NMR experiments (0.7 ms contact time) were carried out using a double-resonance MAS probe with a 4 mm spinning module. Typical 90° pulse lengths were 4 μs for 1H and 13C and heteronuclear decoupling (SPINAL64) at a field strength of ~65 kHz. The 13C chemical shifts were referenced relative to tetramethylsilane. The experiments were conducted at −30 °C and an MAS frequency of 7 kHz. The 13C double quantum 13C single quantum correlation spectra were acquired using the SPC-5 recoupling sequence[48] for double quantum excitation and reconversion (set to 0.571 ms each). The relaxation delay was 2.4 s.

Molecular docking of NPY into Y1R

Peptide docking of full-length porcine NPY was completed using Rosetn class="Chemical">ta’s FlexPepDock application[49]. Briefly, low energy backbone conformations were generated from the starting conformation of UR-MK299-bound Y1R. Initially, the trimer of C-terminal NPY was docked into these conformations using full flexible docking guided by mutagenesis data. For each round of docking 5,000 models were generated. The models were sorted by total energy and binding energy. Top models from a given docking round were used to seed the next round of docking in which the peptide was extended. Fragment picking was performed using the fragment_picker application within Rosetta[26]. Secondary structure during fragment picking was guided by the NMR chemical shift data. Additionally, experimentally derived restraints were used to guide docking (R35-D2876.59, R35-N2836.55, Y36-Y1002.64, R33-N2997.32, L30-I293ECL3). After docking peptides of length 6, 12, 18, and 36, the binding pocket was resampled to allow the ligand binding pocket to adopt to the shape of the peptide. This was accomplished with RosettaCM[50]. The Y1R crystal structure was used as a template along with the docked model to ensure the models did not drift too far from the starting structure though the N terminus was removed until the last docking step to provide steric bulk. Following full-length NPY docking, the N terminus of NPY was localized using loose distance constraints with the peptides identified in crosslinking experiments. Model selection from RosettaCM was accomplished using clustering to ensure backbone diversity. Following the identification of docked poses that satisfied the majority of experimental data, the chemical shifts of docked NPY peptides were calculated and filtered against the experimental NMR data to generate a final ensemble of docked poses with a 1.4 ppm root mean square distance to the experimental data. To analyse the binding interactions, per residue energetic analysis was calculated using the residue_energy_breakdown algorithm. The model with the lowest energy was selected as the representative binding pose (Fig. 4a). The ensemble is rather tight and therefore the individual binding poses are overall similar in structure (Fig. 4b).

Photo-crosslinking experiment between Y1R and NPY N terminus

Cell-free expressed Y1R was produced by coupled in vitro transcription/translation based on the protocol from Schwarz et al.[51] using a bacterial cell n class="Chemical">lysate (S30 extract) from E.coli BL21 (DE3). Soluble membrane protein expression was achieved by addition of 0.1% (w/v) Brij-58, 1 mM GSSG and 5 mM GSH. Expression buffer was then exchanged to a binding buffer (0.1 M Tris/HCl, pH 7.4, 5% glycerol and 0.1% (w/v) Brij-58) and samples were purified by ligand affinity chromatography using [K4[(Ahx)2-biotin]]-NPY immobilized on Pierce Avidin Agarose beads (obtained from Thermo Fisher Scientific, Braunschweig, Germany) as previously described[52]. Elution was performed using 60 mM CaCl2.

For photo-crosslinking Y1R in binding buffer was incubated with [Bpa1,n class="Chemical">K4[(Ahx)2-biotin]]-NPY in a molar ratio of 4:1 (5 nmol:1.25 nmol) for 30 min at room temperature. In addition, the same reaction was performed with an 8 fold-excess of NPY (Y1R:[Bpa1,K4[(Ahx)2-biotin]]-NPY:NPY, 4:1:8). Subsequently, the opened reaction vessels were placed on ice and irradiated with UV light (UV lamp: Atkas Fluorest forte, λ = 366 nm, 180 W) for 90 min. 50 μl of photo-crosslinked Y1R sample (~ 200 μg) was digested with Glu-C and rLys-C (obtained from Promega, Mannheim, Germany) according to the manufacturer’s protocol. Following, crosslinked fragments were isolated by affinity purification using Monomeric Avidin Agarose beads (obtained from Thermo Fisher Scientific, Braunschweig, Germany) according to the manufacturer’s protocol. Possible fragments of digested Y1R were calculated by the online tool PeptideMass[53]. To mind incomplete digestion the tool was allowed for a maximum of five missed cleavages. For the analysis the combined option “Glu C (phosphate) + Lys-C” was chosen. Same was done for the calculation of possible NPY fragments. Theoretical masses of fragments after enzymatic cleavage of photo-crosslinked Y1R – [Bpa1,K4[(Ahx)2-biotin]]NPY were calculated by adding possible Y1R fragment masses to NPY fragment masses containing the N terminus. The masses of Bpa, two times Ahx and biotin reduced by water for formation of a peptide bond were added manually. Peptide fragments of photo-crosslinked Y1R were analyzed by matrix-assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS) using an Ultraflex III MALDI-TOF/TOF mass spectrometer (Bruker Daltonics, Billerica, MA, USA).

Functionality of cell-free expressed Y1R samples was verified by a homogenous binding assay based on fluorescence polarization. We used [Dpr22-atto520]NPY (hereafter: n class="Gene">NPY-atto520) as fluorescence tracer (IP accumulation in transiently transfected COS-7: EC50 = 24 nM, pEC50 = −7.16 ± 0.20). 50 nM of NPY-atto520 was incubated with increasing concentrations of Y1R in Brij-58 micelles in buffer (0.1 M Tris/HCl, pH 7.4, 2.5% glycerol, 0.1% (w/v) Brij-58 and 0.1% bovine serum album) for 90 min under gentle agitation in opaque 96 well plates. Fluorescence was then measured in a Tecan Spark plate reader (Tecan, Männedorf, Switzerland) using linear polarized light, excitation 510/5 nm, emission 550/10 nm, 90° detection angle. Experiments were conducted at least twice independently in duplicate.

Data availability

Atomic coordinates and structure factor files for the Y1R–n class="Chemical">UR-MK299 and Y1R–BMS-193885 complex structures have been deposited in the Protein Data Bank with identification codes 5ZBQ and 5ZBH, respectively.

a, b, Crystal packing of Y1R–n class="Chemical">UR-MK299 (a) and Y1R–BMS-193885 (b) complexes. Y1R is shown in cartoon representation and coloured brown and green in the Y1R–UR-MK299 and Y1R–BMS-193885 complexes, respectively. The T4L fusion is shown in grey cartoon representation. UR-MK299 and BMS-193885 are displayed as yellow and pink spheres, respectively. c, Cutaway view of UR-MK299 binding pocket in Y1R. The receptor is shown in brown cartoon and surface representations. The ligand is shown as yellow sticks. d, Comparison of Y1R between the Y1R–UR-MK299 crystal structure (brown) and Y1R–NPY model (green). Side chains of Q1203.32 and W2766.48 are shown as sticks. R35-Y36 of NPY is displayed as cyan sticks. The hydrogen bond between Q1203.32 and Y36 of NPY is shown as a green dashed line. e-j, Chemical structures of argininamide Y1R antagonists, BIBP3226 (e), UR-HU404 (f), UR-MK299 (g), BIBO3304 (h), UR-MK289 (i) and UR-MK136 (j). k, Chemical structure of BMS-193885. l, Scaffold of NPY C-terminal residues R35 and Y36. Key differences between R35-Y36 of NPY and UR-MK299 are chirality of the arginine derivative and alteration of bond connectivity leading to the hydroxyphenyl group.

a, Live cell fluorescence microscopy verifies all Y1R variants to be properly folded and exported to the cell membrane like the wild-type receptor. Nuclei stained with n class="Chemical">Hoechst33342, bar equals 10 μm. Pictures are representative of two independent experiments with similar results. b, Total expression level was determined by fluorescence reading and confirmed expression similar to the wild type. Transfection of only 50% or 25% of the DNA amount (with total DNA amount held constant by empty vector), respectively, led to a proportial decrease of fluorescence, and thus, expression level. Data represent means ± s.e.m of three to five independent experiments performed in technical triplicate (see source data for sample size n of each mutant). c, Estimation of the receptor reserve in functional IP accumulation assays. Transfection of half of the vector encoding receptor (with constant total DNA amount including chimeric G protein, cf. a) still produces maximum signal, while further reduction results in signal loss at comparable potency. Thus, there is only a small receptor reserve in the functional readout, allowing potency alteration to be directly related to compromised ligand binding. Data represent means ± s.e.m of three independent experiments performed in technical duplicate.

Colours represent similarities of residues: red background, identical; red, strongly similar. Key residues in the n class="Chemical">UR-MK299 binding pocket, which are conserved or variable among receptors, are indicated by red and black arrows, respectively. The alignment was generated by uniprot (http://www.uniprot.org/align/) and graphic was prepared on ESPript 3.0 server (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi).

a, Binding of atto520-labelled NPY (50 nM) to increasing amounts of Y1R-conn class="Chemical">taining bicelles or empty bicelles. Data reflect fluorescence enhancement upon binding. An inflection point at EC50 = 52 nM was determined. Two independent experiments were performed in technical duplicate with similar results. Data shown are from a representative experiment. b, Typical 13C MAS single quantum (SQ) / double quantum (DQ) correlation spectrum of NPY in the presence of Y1R reconstituted into large bicelles at −30 °C. NMR spectra were acquired from 1 to 3 independent preparations for each labelled amino acid with similar results (see panel d). Data shown are from a representative experiment. c, Table showing 13C NMR chemical shifts of assigned amino acids of NPY bound to Y1R (referenced to tetramethylsilane) as acquired in solid-state NMR experiments. d, 13C chemical-shift index of NPY bound to Y1R in large DMPC/DHPC-c7 bicelles (q > 20) compared with docked models. Plotted is the measured chemical shift difference (Cα-Cβ) for each individual residue of NPY subtracted by chemical shift difference of the same amino acid type in random coil conformation in black. Individual data points from 1 to 3 independent experiments for each labelled amino acid are shown. Typical experimental errors in determining chemical shifts under these conditions are ±1 ppm. Chemical shifts were back-calculated for the top docking solutions and filtered against the experimental data to generate a final ensemble of docked poses. Their average chemical-shift index and associated standard deviation from 10 top docked poses are shown in red.

a, Mass spectra of photo-crosslinked Y1R with [Bpa1,n class="Chemical">K4[(Ahx)2-biotin]]NPY. Exemplary MALDI-ToF mass spectra of photo-crosslinked samples enzymatically digested by rLys-C and Glu-C. Potential Y1R fragments are labelled. Two independent experiments were performed with similar results. N, N terminus of Y1R (blue); E, ECL2 (red). b, Respective regions of NPY N terminus at Y1R. Amino acid sequence of Y1R with a C-terminal His-tag. The two detected regions within Y1R (N terminus (blue), ECL2 (red)) after crosslinking with [Bpa1,K4[(Ahx)2-biotin]]NPY are emphasized in boxes. The different sizes of the boxes represent different detected fragments (Extended Data Table 5), respectively. Experiments were repeated twice independently with similar results, and only fragements that were observed in both experiments are listed here and in Extended Data Table 5. c, Binding of atto520-labelled NPY (50 nM) to increasing amounts of cell-free produced Y1R in Brij-58. Data reflect fluorescence enhancement upon binding. EC50 value of 69 nM was determined. Data shown are means ± s.e.m. from six independent experiments performed in technical triplicate.

EC50 values were determined after 1 h stimulation by increasing concentration of NPY or n class="Gene">NPY together with different antagonists. Data are shown as means from at least three independent experiments (n>2) or results of two individual experiments (n=2) each performed in technical duplicate.

The EC50 ratio represents the shift between the NPY and n class="Gene">NPY/antagonist curve (EC50(NPY+antagonist) / EC50(NPY)) and characterizes the antagonistic effect on wild-type receptor or receptor mutants. By comparison of EC50 ratios between wild type and receptor mutant, influences of all tested residues on antagonist activity were determined. The higher ratio, the higher antagonist activity. A reduced EC50 ratio of mutant compared to the wild-type receptor was interpreted as important for the respective antagonist.

These data were obn class="Chemical">tained at a reduced concentration of UR-HU404 (10−8 M) as concentration response curves did not reach saturation (EC50 > 10,000 nM) when high concentration was used (10−7 M).

Dissociation constant determined by satn class="Chemical">uration binding at Sf9 membranes (receptor expression was confirmed by western blot analysis) using a sodium-containing buffer (a) or a sodium-free buffer (b) (Note: the sodium-free buffer was used for the determination of agonist binding affinity because porcine NPY exhibited approx. 10-fold higher affinity in the sodium-free buffer compared to the sodium-containing buffer (data not shown)).

Dissociation constant determined by competition binding with [n class="Chemical">3H]-UR-MK299 at Sf9 membranes using a sodium-containing buffer (a) or a sodium-free buffer (b).

EC50 were determined using GraphPad Prism 5.0. All curves were normalized to the top and bottom values of the n class="Gene">Y1R/NPY curve. Nonlinear regression (curve fit) was performed for normalized response in all assays. All data are shown as means from at least three independent experiments (n>2) or results of two individual experiments (n=2) each performed in technical duplicate.

The EC50 shifts were determined by EC50(mutant) / EC50(WT), set Hill slope to 1. For the wild-type receptor x-fold is set to 1. Lower EC50 shift of n class="Gene">NPY analogue/mutant compared to NPY/mutant was interpreted as no further loss of function and a direct interaction between both positions.

Selected calculated masses in Dalton of possible photo-crosslinked fragments of Y1R with [Bpa1,n class="Chemical">K4[(Ahx)2-biotin]]NPY or [Bpa1,K4[(Ahx)2-biotin]]NPY with itself. The fragments are selected based on the correlation with the detected signals. For clarity, further calculated masses of possible photo-crosslinked fragments are not shown.